Using Machine Learning for Automatic Estimation of M. Smegmatis Cell Count from Fluorescence Microscopy Images

Vente, Sam; Arandjelović, Ognjen; Baron, Vincent O.; Dombay, Evelin; Gillespie, Stephen H.

doi:10.1007/978-3-030-24409-5_6

Cited by 7 publications

(10 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The best performance was obtained using the simplest and shallowest model, namely ResNet18. Its error of less than 5% is a significant improvement on all previous work, therefore we likewise note here the attainment of the new state of the art [12,21,22]. The visualizations shown in Figure 7 provide further insight into the learning achieved using ResNet18.…”

Section: Bacterial Countsupporting

confidence: 56%

“…Thus, the predicted pseudo-count of 1.05 can be interpreted as more confidently corresponding to a single bacterium than, say, 1.48 (whereas 1.51 would tilt the decision towards the count of 2). Our approach also allows for the cancellation of uncorrelated errors across the slide, as observed in previous research [12].…”

Section: Bacterial Countmentioning

confidence: 94%

“…Unfortunately, the method was analysed rather poorly, without any error analysis being reported. In one of the most recently published papers on the topic, Vente et al proposed a fairly complex method for the localization of bacteria, employing edge detection, Fourier analysis, and morphological operators [12], and thereafter estimating the bacterial count in the regions of interest using simple regression. The authors reported an error of 6.5% in the empirical experiments.…”

Section: Related Workmentioning

confidence: 99%

“…Researchers who study the metabolic adaptation of M. tuberculosis to drug pressure and other physiological stresses are also interested in the microscopic appearances of individual bacterial cells. Changes in the length and width of cells [10], loss of acid fastness [11], and accumulation of intracellular lipid [11][12][13] may influence the transmissibility and antibiotic tolerance of M. tuberculosis. Microscopy remains an important tool for the description and investigation of these features.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Tuberculosis Bacteria Detection and Counting in Fluorescence Microscopy Images Using a Multi-Stage Deep Learning Pipeline

et al. 2022

Self Cite

View full text Add to dashboard Cite

The manual observation of sputum smears by fluorescence microscopy for the diagnosis and treatment monitoring of patients with tuberculosis (TB) is a laborious and subjective task. In this work, we introduce an automatic pipeline which employs a novel deep learning-based approach to rapidly detect Mycobacterium tuberculosis (Mtb) organisms in sputum samples and thus quantify the burden of the disease. Fluorescence microscopy images are used as input in a series of networks, which ultimately produces a final count of present bacteria more quickly and consistently than manual analysis by healthcare workers. The pipeline consists of four stages: annotation by cycle-consistent generative adversarial networks (GANs), extraction of salient image patches, classification of the extracted patches, and finally, regression to yield the final bacteria count. We empirically evaluate the individual stages of the pipeline as well as perform a unified evaluation on previously unseen data that were given ground-truth labels by an experienced microscopist. We show that with no human intervention, the pipeline can provide the bacterial count for a sample of images with an error of less than 5%.

show abstract

Section: Bacterial Countsupporting

confidence: 56%

Section: Bacterial Countmentioning

confidence: 94%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Tuberculosis Bacteria Detection and Counting in Fluorescence Microscopy Images Using a Multi-Stage Deep Learning Pipeline

et al. 2022

Self Cite

View full text Add to dashboard Cite

show abstract

“…Recently, artificial intelligence methods such as machine learning have been used not only to estimate the bioactivity of drug candidates [ 176 ] but also to complement existing tools across different stages in the NTM drug discovery process, for example, in automating the cell count from fluorescence microscopy imaging and identification of mycobacteria species from mass spectroscopy [ 177 , 178 ].…”

Section: Models For Drug Discovery Against Ntmmentioning

confidence: 99%

Drug Resistance in Nontuberculous Mycobacteria: Mechanisms and Models

Saxena

Spaink

Forn-Cuní

2021

Biology

View full text Add to dashboard Cite

The genus Mycobacteria comprises a multitude of species known to cause serious disease in humans, including Mycobacterium tuberculosis and M. leprae, the responsible agents for tuberculosis and leprosy, respectively. In addition, there is a worldwide spike in the number of infections caused by a mixed group of species such as the M. avium, M. abscessus and M. ulcerans complexes, collectively called nontuberculous mycobacteria (NTMs). The situation is forecasted to worsen because, like tuberculosis, NTMs either naturally possess or are developing high resistance against conventional antibiotics. It is, therefore, important to implement and develop models that allow us to effectively examine the fundamental questions of NTM virulence, as well as to apply them for the discovery of new and improved therapies. This literature review will focus on the known molecular mechanisms behind drug resistance in NTM and the current models that may be used to test new effective antimicrobial therapies.

show abstract

From Precision Medicine to Precision Health: A Full Angle from Diagnosis to Treatment and Prevention

Shaban-Nejad

Michalowski

2019

Precision Health and Medicine

View full text Add to dashboard Cite

Using Machine Learning for Automatic Estimation of M. Smegmatis Cell Count from Fluorescence Microscopy Images

Cited by 7 publications

References 23 publications

Tuberculosis Bacteria Detection and Counting in Fluorescence Microscopy Images Using a Multi-Stage Deep Learning Pipeline

Tuberculosis Bacteria Detection and Counting in Fluorescence Microscopy Images Using a Multi-Stage Deep Learning Pipeline

Drug Resistance in Nontuberculous Mycobacteria: Mechanisms and Models

From Precision Medicine to Precision Health: A Full Angle from Diagnosis to Treatment and Prevention

Contact Info

Product

Resources

About